Prepare for a decision you will have to defend
The meeting does not measure how much material you brought. It measures three things: whether you know what other options existed, what would make your choice fail, and what you still do not know.
- 01 · Master a new subject
- 02 · Absorb documentation
- 03 · You are here
- 04 · Cut task time
Build a one-page brief with four things: the decision, the options you ruled out, three reasons this could go wrong (with the first sign of each), and the doubt you state yourself before anyone asks.
Sign up once to unlock this course
Course 01 is open. For courses 02 to 04 we ask for six fields. It is a single signup: done once, valid for the whole Academy. It is not a free diagnosis and it does not trigger automatic sales contact.
Arriving with material is not arriving prepared
The meeting where you defend a recommendation does not measure how many pages you brought. It measures three things: whether you know what other options existed, whether you know what would make your choice fail, and whether you know what you still do not know.
An AI tool assembles the material in minutes. None of them delivers those three things, because all three depend on you carrying the consequence of the choice.
Imagining it already went wrong changes what you can see
In 2007, in the Harvard Business Review, the psychologist Gary Klein proposed a very simple procedure. Instead of asking “what could go wrong?”, the group starts from the assumption that the project has already failed and writes down the reasons for the failure. He called it a premortem.
The difference is not just a turn of phrase. Faced with a failure presented as fact, a person stops estimating probability and starts explaining an event, which is far easier and far more detailed. Klein cites 1989 research by Deborah Mitchell, Jay Russo and Nancy Pennington finding that imagining something has already happened increases by 30% the ability to correctly identify the reasons for a future outcome.
AI helps well here: describe your recommendation and ask for the reasons it failed six months later. What it does not do is decide which of those risks you are willing to run. That is exactly the part the meeting will hold you to.
One page, four blocks
The whole preparation fits on one page. What does not fit on one page is usually not decided — just accumulated.
| Block | What to write | What AI can do |
|---|---|---|
| 1. The decision | One sentence starting with a verb, fitting on one line. Not the topic: the choice. “Change the approval flow from X to Y.” | Nothing. If AI writes that sentence, the decision is not yours. |
| 2. The options you ruled out | Two or three, each with a one-line reason. Always include “do nothing”, which is a real option. | Suggest options you did not think of, for you to rule out with a reason of your own. |
| 3. How this failed | Three reasons this went wrong six months from now, and what the first sign of each would be. | List failure reasons from your description, including the uncomfortable ones. This is the best use of the tool on this page. |
| 4. The doubt that remains | What you do not know that would change your recommendation. Written before the meeting, not made up on the spot. | Point out the holes in your argument, which usually surfaces the question you were avoiding. |
Read your page aloud thinking about the most sceptical person in the room. If the objection that comes into your head is not in block 3 or block 4, the page is not ready yet.
Presenting as settled what nobody verified
The most common mistake is not recommending the wrong thing. It is presenting as settled a point nobody checked, and being found out in front of everyone.
The recommendation that fell on the third question
A coordinator proposes changing an approval flow. The material is impeccable: context, numbers, comparison with the current process, a timeline. The tool helped organise all of it in a single afternoon.
The third question in the meeting is about the cases that are handled by hand today, outside the normal flow. He cannot answer: no summary raised it, and he had not written his doubt down. The proposal was not rejected — it was postponed, which costs more.
The question that changes what you can see
Both conversations use almost the same words and produce very different lists. The only difference is the tense of the verb.
“What could go wrong” against “it went wrong, why”
The first question makes everyone defend the proposal. The second starts from the assumption that it already went wrong, and so forces people to explain instead of reassure.
What could go wrong?
- Whoever proposed it answers by defending the proposal.
- The risks that come up are the ones that already have an answer ready.
- Disagreeing sounds like distrust of the plan, so almost nobody disagrees.
- The usual result: a short, already familiar list.
We are six months ahead and this failed. Why?
- The failure has already been stated as fact, so there is nothing to defend.
- The task becomes explaining, and explaining requires detail.
- Naming a reason stops looking like disloyalty and becomes what was asked for.
- The result: specific reasons that the first question never reaches.
Present the decision as already taken and already failed, with a date. “It is February and this did not work.”
Alone and on paper, before any group discussion.
On one side, what you can prevent now. On the other, what you can only keep an eye on.
Why does this question surface more reasons than the usual one?
What does the meeting actually test?
What is the most common mistake in the meeting?
Try to answer from memory before you click. Same as course 01: trying to remember makes it stick, rereading only raises your confidence.
Where the argument comes from
- Gary Klein, “Performing a Project Premortem”, Harvard Business Review, vol. 85, no. 9, September 2007, pp. 18–19, reprint F0709A. What the article says: in a premortem, team members assume the project they are planning has just failed and then list plausible reasons for that, which gives people permission to raise their doubts beforehand rather than afterwards. The 30% increase in correctly identifying reasons for future outcomes is attributed by Klein to 1989 research by Deborah J. Mitchell, Jay Russo and Nancy Pennington. It is a figure Klein quotes from other research, not one he measured himself. We record it that way on purpose. Verified 6 Aug 2026: authorship, journal, issue, month, pages and reprint code confirmed on HBR's page, the HBR store and two independent bibliographic references. Divergence recorded: a second-hand source claims the method “doubles” the number of risks found. That figure is not in Klein's article and was discarded.
The next course
The first three courses are about studying, reading and deciding. Course 04 closes the track on time: how to cut your own, without confusing a real gain with one that only looks like a gain.